Robot Instruction through Bayesian Approaches to Laban-based Manipulative Action Characterization
|Project Type: PhD Project|
|Research Field: Human and Robotic Dexterous Manipulation|
|Time span: 12/2010-6/2015|
This project seeks to develop a hierarchical Bayesian framework to model human-like grasp behaviors and corresponding action sequences given labeled data resulting from LHMA (Laban-based Human Motion Analysis) components. This framework will allow a robotic system to learn by demonstration how to classify the different grasps employed to manipulate different objects and consequently reproduce them using computer vision and an artificial hand.